The responsible admin approves.
- 01 / Review the exact change and target.
- 02 / Inspect impact, risk, and missing evidence.
- 03 / Approve or reject the specific proposal.
Pending or rejected? Execution stays blocked.
OneAdmin is built around an admin inbox and an explicit execution gate: human approval or a previously assigned policy. Agent-triggered work stays visible, from proposal to result.
Walk through the control modelChoose a proposal. Follow its gate.
Make the decision in the inbox below.
BI-104 The agent has proposed a change. Without approval or an applicable assigned policy, it stops here.
Finance requested a percentage variance. The agent proposes the expression below for administrator review.
SUM(Finance[Actual]) - SUM(Finance[Budget])
DIVIDE( SUM(Finance[Actual]) - SUM(Finance[Budget]), SUM(Finance[Budget]) )
The agent can submit this proposal. It cannot execute without your decision or an applicable assigned policy.
Demonstrates the hold. No approval or assigned policy means no execution.
The intended contract is the same for every initiator: agent, administrator, or integration. A trigger creates a request. It does not grant permission to execute.
TARGET CONTROL MODEL · UNIFIED ENFORCEMENT PLANNEDApproval and policy both operate within an authorized scope. Neither expands the actor’s underlying permissions.
Pending or rejected? Execution stays blocked.
Missing, expired, or out of scope? Back to the admin.
Policy-covered work remains in the admin’s view with the policy version and assignment. Human-approved work retains its reviewer. Blocked attempts and rejections remain visible too.
A change starts in one place. Its consequences can travel through planning models, calculations, reports, and teams.
The approval should carry that evidence: what changes, what depends on it, and which impacts remain unverified.
Analytics operations
The admitted model provides the structure for inspecting calculations and supported column dependencies.
*The cross-system map is illustrative. Current Power BI dependency coverage starts at measure-to-column relationships; enterprise asset lineage is planned.
Explore the lineage roadmapPlatform-specific agents and tools are the design foundation. The intended shared gate brings each proposal into the responsible admin’s view, with its authority and recorded outcome.
Models, measures, workspaces
Application snapshots & artifacts
INITIAL IMPLEMENTATION FOCUS
Agent proposal.
Approval or assigned policy.
A retained decision record.
Manual requests & generic webhooks
Review, authorize, and follow the outcome
ServiceNow, Jira, and Slack provider workflows are planned extensions.
Tableau / Looker / SAP BPC
Future coverage| Capability | Power BI | Oracle EPBCS |
|---|---|---|
| Inventory & inspection | Admitted models and workspaces | Uploaded application snapshots |
| Dependency context | Measure-to-column dependencies | Parsed application artifacts |
| Change preparation | Selected manual request workflows | Snapshot-based investigation |
| Live platform actions | Dedicated publishing workflow | Planned |
| Cross-system lineage | Planned | Planned |
| Unified approve & execute | Planned | Planned |
Implementation scope varies by platform and access method. This is not a general availability commitment.
Walk through an admin inbox, review an agent proposal, and test what happens when approval or assigned policy is missing. Discuss the enforcement scope your deployment needs.
Let’s talk operations